What Mechanisms Drive the Long-run Effects of Career and Technical Education? Evidence from Connecticut

• In Event: Understanding the Long-Run Effects of Career and Technical Education: An Examination of Impact, Mechanism, and Implementation

Abstract

Career and Technical Education (CTE) programs strive to support the development of students’ skills for career success. Although past work has demonstrated that CTE can produce long-run benefits on important adult outcomes, such as earnings (Brunner et al., 2023; Lindsay et al., 2024), fundamental questions remain about how CTE shapes students’ long-run trajectories. Indeed, there have long been concerns that some of the short-term benefits of CTE may come at the cost of long-term disadvantages (Dougherty, 2021; Hanushek et al., 2017; Krueger & Kumar, 2004). Although researchers have attempted to probe the plausibility of different candidate mechanisms – for example, short-run gains in general academic skills or engagement – it is challenging to isolate causal processes in the context of a single study.

Using data from the Connecticut Technical High School System, the current study aims to advance our understanding of the mechanisms that drive CTE’s long-run effects by examining the extent to which the initial effects of CTE on student test scores, high school graduation, and attendance forecast the long-run impacts of CTE on earnings and educational outcomes.

The current study builds directly on past work that has estimated the causal effects of 16 oversubscribed CTE high schools in Connecticut across 8 cohorts using a regression discontinuity design (Brunner et al., 2022, 2023, 2025). In this study, however, we use regression discontinuity methods to compute the longitudinal effects of CTE for each school by cohort cluster (k = 128). In effect, we will treat each cluster as an individual CTE evaluation. We will estimate short-run effects on test scores, attendance, and high school graduation and long-run effects on annual earnings measured 4 to 5 years after graduation and educational outcomes including semesters of college attendance and degree/certificate achievement, as measured in the 1 to 5 years after on-time high school graduation.

To examine whether short-term effects on theoretically relevant outcomes correspond with long-run effects, we will run meta-regression models in which we will regress cluster-level follow-up impacts on educational outcomes and earnings on short-run impacts. We will weight by the inverse variances of the follow-up effects and include a random effect for school to address non-independence.

As a first step in our analysis, we examined whether there was meaningful heterogeneity in cluster-level effects for short-term outcomes. We observed statistically significant Q statistics and non-zero tau values (average tau = .09) suggesting that there is, indeed, detectable and meaningful variation in the effect of CTE offers across clusters. Clusters that generated larger effects on math tended to generate larger effects on reading and attendance (r ~ .40), but not graduation (r ~ 0). Effects on attendance and graduation were correlated at r = .32.

Our preliminary analyses indicated meaningful variation in the effects of CTE across clusters. Prior to the conference, we will execute the proposed predictive models. We anticipate that our analyses will improve our understanding of the mechanisms that may drive the long-run effects of CTE, with implications for our broader understanding of human capital formation writ large.

Author

• Emma Hart, Boston College
  Presenting Author